Computer Vision

The state of the art technology in the Aveon® customer engagement system uses overhead video cameras to feed images back to a computer. Images are processed by an advanced tracking algorithm to record people movements within the picture frame.

The ‘complete’ path followed by a person is recorded, from the moment they appear till they exit the field of view. Counts are generated when a person's path crosses a counting zone drawn on the image. Each count has associated with it the direction of travel and time of occurrence. Because all people movements within the image are tracked, multiple zones can be used in a single camera view. The data from multiple cameras can be tied together in real-time, without duplication, to handle entrances too wide to be covered by a single camera, or to provide coverage of a broad area.

Data Visualisation

The Aveon® system uses advanced data visualisation technology to provide a virtual view of a property, whether in-store layout with merchandising and queuing, or physical casino/mall with entrance traffic and retailer. Multiple datasets can be visualised across a business's whole operation in one view.  Data sets can be ranked by colour to display relative performance, or use of heat maps to illustrate movement or dwell.

98% Accuracy Guarantee

This unique process addresses a large number of accuracy problems commonly encountered in alternative systems.  These include variable lighting conditions, shadowing, different camera heights and people sizes, high traffic volumes and people loitering or pausing.  We are the only company in the world that has the capability and the confidence to provide a guarantee of 98% accuracy.

When evaluating the accuracy of a system it is important to measure accuracy over time periods no larger than the minimum used for any application of the data. For example, for a rostering system using hourly binned data for planning, accuracy needs to be evaluated over hourly bins or smaller.  This is because a given accuracy level over a longer period does not guarantee that the same accuracy level is achieved over the smaller periods therein.  A system may over count in the mornings and under count in the afternoons but still appear accurate over the whole day.